Welcome

I’m Carlos, a Principal DevOps / SRE Engineer based in Donostia (San Sebastián), in the Basque Country.

I build and maintain the infrastructure that keeps production systems running at scale — Kubernetes clusters, cloud platforms on AWS, CI/CD pipelines, and the observability layers that make it all visible.

My daily work spans EKS, Terraform, Helm, GitLab CI/CD, and increasingly, the infrastructure behind AI/ML workloads: vLLM, KServe, and GPU-aware scheduling.

The Homelab as a Career Pillar

Most engineers have a side project. I have a production environment at home.

My homelab is a self-hosted Proxmox cluster with GPU nodes, running Talos-based Kubernetes with full observability — Prometheus, Grafana, Loki, Tempo, and OpenTelemetry. It’s not a curiosity or a weekend distraction. It’s where I validate infrastructure patterns against real hardware, real constraints, and real failure modes before they ever touch a production environment.

This is where I’ve tested Karpenter scaling strategies, experimented with LLM inference optimization on actual GPUs, debugged distributed tracing pipelines end to end, and broken things in ways that made me a better engineer at work. The philosophy is simple: break things at home so you don’t break them where it matters.

What You’ll Find Here

This portfolio is a window into that practice — professional work, homelab experiments, and the intersection of infrastructure engineering with AI/ML workloads. Expect writing on Kubernetes internals, operator patterns, MLOps on real hardware, and the kind of observability insights that only come from running your own stack.

If any of this resonates, feel free to explore the projects and posts, or reach out directly. I’m always happy to talk shop.

— Carlos